SPEAKER POSITION DETECTOR

摘要

PROBLEM TO BE SOLVED: To highly accurately detect a speaker position by judging it while adding two information of the detection information of a sound source position and the detection information of a figure position. SOLUTION: A sound source position detection part 3-1 of a control part 3 prepares a sound source position may by inputting sound source information inputted from a microphone array 1. Based on the image information inputted by a sensor 2 for image input, a human body position detection part 3-2 prepares a human body position map. The map calculates the probability for the sound source or human body position to exist for the unit of each domain while partitioning the range of a space detectable for the microphone array 1 and sensor 2. A speaker position discrimination part 3-3 calculates the product of probabilities in the correspondent domains of the sound source and human body position maps and discriminates the domain having the largest product as the speaker position. Either one of an ultrasonic sensor, an infrared sensor or a telecision camera is used for the sensor 2.